Temperature is one of the key environmental factors affecting flock health, thermal comfort, and production in poultry houses. This study aims to clarify the spatial distribution of temperature data in large-scale poultry houses using an inspection robot. Environmental data was collected by inspection robots. The experiment was conducted in a commercial multi-tier caged breeder poultry house in Shijiazhuang, Hebei Province, China. The poultry house with the size of 96 m × 12 m was oriented from east to west. A four-row and five-aisle layout was adopted with two floors and three cage tiers on each floor, resulting in a total of six tiers. 40000 birds were housed in one building. The poultry house was equipped with a pad-fan ventilation and cooling system, sidewall ventilation windows, and an environmental control system. The inspection robot was developed with a mobile platform, a multi-parameter environmental monitoring module, an image acquisition module, and a data system. According to the production schedule, the robot inspected the first floor twice a day. In winter, inspections were carried out at 08:30 and 11:00, while in summer they were carried out at 09:00 and 11:00. The robot moved along an S-shaped route at a speed of 0.2 m/s, and one full inspection of the poultry house took about 40 min. Outdoor temperature was recorded by a HOBO sensor installed outside the building. Ventilation conditions were obtained from the environmental controller. Temperature data were collected over ten days in winter and ten days in summer. Before spatial analysis, the data were preprocessed using cleaning and anomaly detection. Invalid, duplicated, and abnormal records were removed or corrected, and an Isolation Forest model was applied to identify anomalous observations. Global spatial autocorrelation analysis using Moran's I was used to evaluate the overall spatial dependence of temperature. Local spatial autocorrelation analysis using the Getis–Ord G statistics was applied to identify temperature hot and cold spots in different regions of the poultry house. The potential causes of seasonal temperature heterogeneity were explained to evaluate spatial analysis with outdoor temperature, indoor temperature, ventilation level, and building layout. The results showed that the temperature distribution in the poultry house shared significant positive spatial autocorrelation in winter and summer. The Moran's I values ranged from 0.295 to 0.631 in winter and from 0.138 to 0.616 in summer, and all inspection periods reached a significant level (P = 0.001), indicating the temperature distribution with different degrees of spatial clustering. In winter, temperature generally presented stronger spatial aggregation, and the clustering intensity was more strongly dominated by outdoor temperature and the indoor–outdoor temperature difference. Ventilation regulation served to maintain air quality under low ventilation levels, indicating weak temperature clustering. In summer, the spatial clustering pattern of temperature also changed with inspection time and date. In most cases, the clustering intensity in the morning was higher than that at noon, because high-level ventilation with evaporative pad cooling promoted air exchange and heat removal, thereby reducing temperature aggregation. Once the outdoor temperature was high, the cooling effect of the evaporative pad system was weakened due to temperature clustering. Local spatial autocorrelation analysis further revealed that there was outstanding spatial heterogeneity inside the poultry house. In winter and summer, cold spots were distributed near the external walls, entrance, sidewall inlets, and evaporative cooling pad end, whereas hot spots were concentrated in the middle aisles and central regions. The central area was more likely to accumulate heat, due to flock heat production, weak heat loss, and limited local air mixing. In winter, the proportion of hot spots generally fluctuated between 40% and 50%, while the proportion of cold spots remained around 25% to 35%. In summer, the proportions of hot and cold spots fluctuated more strongly, but the occurrence frequency of local hot spots was generally higher than that of cold spots in both seasons. These findings indicate that temperature in a large-scale multi-tier caged poultry house represented significant spatial heterogeneity and seasonal clustering. Winter temperature clustering was closely related to outdoor temperature and heat loss from peripheral areas, whereas summer clustering was combined to regulate high ventilation and evaporative cooling. The findings can provide a data basis to optimize building insulation, ventilation, and cooling strategies in poultry houses using inspection robots.